Breakdown field comparison for different plastics

Apart from PA-66, there are several other plastics which have been used as enclosure materials in experiments and numerical models. In this subsection, we demonstrate the capability of our kinetics solver to compare the performances of different plastics by plotting their respective breakdown fields. From the Ecrit values for pure PA-66,

PETP and PMMA plotted in figure 6.6, it can be observed that PA-66 possesses higher breakdown fields compared to PETP and PMMA and hence, is preferable to the other plastics. While this observation is in agreement with published data [13], the current work represents the first effort to quantitatively compare the performances of plastics

in terms of their breakdown field strengths. Previous works have argued for the better performance of PA-66 on the basis of thermodynamic (Cp) or transport (σ) properties,

rather than a dielectric property (Ecrit). The primary reason for better performance

of PA-66 is that among the plastics under consideration, PA-66 is the only plastic which contains nitrogen atoms and this translates to greater HCN number densities. Inclusion of the contribution from CN is expected to significantly increase the gap in Ecrit between PA-66 and the other two plastics. Since the effects of C(gr) are not

included in the EEPF and Ecrit calculations, the results in figure 6.6 are not expected

to be accurate below 3,000 K. Nonetheless, the profile observed below 3,000 K is physically acceptable, since it has been established experimentally [114, 115] that the metallic particulate contamination reduces the breakdown threshold for atmospheric pressure air gaps.

Chapter 7

Comparison with experiments

In the previous chapters, we focused on determining dielectric breakdown fields for general air-metal-plastic vapor mixtures and presented results for air-copper-PA66 mix- tures. We now focus on utilizing the breakdown field results towards predicting dielec- tric breakdown for a known circuit breaker geometry, for which experimental results on breakdown are available post-CZ. In the pre-CZ high-current phase, the arc volt- age results from the CFD solver [14] for a free burning arc were observed to be in good agreement with experiments. As a logical extension, the CFD solver was deemed appro-

Figure 7.1: Experimental set-up with stationary electrodes and the corresponding com- putational mesh for a 2D axi-symmetric geometry [14].

priate to analyze fluid flow post-CZ and provide the required input data for the kinetics solver developed in current work. In addition to the CFD solver, R¨umpler and his co- workers have devised experimental set-ups for studying dielectric breakdown post-CZ. Experimental results are available for a simple circuit breaker geometry shown in figure 7.1, involving air-copper mixtures without plastic wall ablation. The details regarding the experiments are provided in R¨umpler’s doctoral thesis [14]. In the experiments, a well-defined voltage pulse is applied across the contact gap after CZ and time at which

Figure 7.2: Experimental results [14] at different delay times (tv) for the (a) Reignition

and (b) Extinction cases. The delay times (in µs) post-CZ for the experiments numbered 1-4 in each case are given in table 7.1.

Table 7.1: Delay times (µs) for the experimental data in figure 7.2 for (a) Re-ignition and (b) Extinction cases.

Expt. no. Re-ignition Breakdown Extinction Breakdown

1 44.3 Yes 80.9 No

2 44.0 Yes 128.5 No

3 100.0 Yes 93.4 No

the spark breakdown occurs post-CZ is noted. The load voltages (Ul) across the contact

gap are generated by the Surge Impulse Generator (SIP). Accordingly, numerical simu- lation data from a CFD set-up which closely mimics the experimental set-up is required and as a consequence, the same voltage pulse applied on the gap in the experiments is also applied after a suitable time delay as nodal voltage. The computational mesh for the CFD simulation’s 2D axi-symmetric geometry is also presented in figure 7.1. In this chapter, we first briefly describe the numerical procedure utilized to generate CFD data for an equivalent experimental set-up. Elaborate descriptions of the numerical proce- dure are available in R¨umpler’s thesis [14]. Of particular interest to the current work, we then describe a simple methodology to predict dielectric breakdown for a given CFD data post-CZ and compare the numerical predictions with the experimental results.

7.1

Experimental results

Experimental results are available from R¨umpler [14] for two different load voltage magnitudes: (i) Ul = 1.2 kV , and (ii) Ul = 2.0 kV , labeled as the “Extinction” and

“Re-ignition” cases respectively for convenience. The terminology implies a greater pos- sibility of dielectric breakdown for the “Re-ignition” case owing to the greater magnitude of the load voltage. For both the cases, the load voltages are applied after different delay times (tv) post-CZ in order to allow for different levels of arc cooling. In general, we

expect the arc temperatures to decrease with increase in the delay time.

We are particularly interested in the arc voltage profiles across the contact gap, which are provided in figure 7.2. The delay times post-CZ for application of the TRV in both cases in listed in table 7.1. Breakdown of the contact gap is characterized by the sudden collapse of arc voltage, and the time between CZ and the collapse of the arc voltage is termed the “breakdown time” (tbr) measured in µs. We first consider the

re-ignition case. From the plots in figure 7.2(a) and data in table 7.1, it can be clearly seen that breakdown is observed for delay times lower than 100 µs while no breakdown is observed for the delay time of 128 µs. Also, it can be straight-forwardly deduced that the breakdown times (tbr) increase with increase in delay times (tv). Similarly, for the

extinction case in figure 7.2(b), no breakdown is observed for any of the experimental trials and the minimum delay time for which breakdown does not occur is 67.2 µs.

Figure 7.3: Electrical circuit including a Surge Impulse Generator (SIP) source with a parallel capacitor to generate the voltage pulse after CZ [14].

Apart from arc voltages, several transport variables are required for characterizing dielectric breakdown in the current work. It is important to note that the transport variables of interest to current work are difficult to be obtained experimentally for the entire domain, owing to the miniature contact gap. The CFD solver is expected to resolve this issue. The accuracy of the CFD solver can also be quantified by the proximity of its arc voltage results to the experimental results. The numerical arc voltage results will be presented in the next section and reasonable agreement with experimental results will be observed. Hence, as an extrapolation, the transport variables from the CFD solver are expected to closely resemble the corresponding experimental results.